Re: [PATCH v2 3/4] Device bindings documentation updated ACPI-enabled platforms not currently supported

From: Ramiro Oliveira
Date: Tue Oct 18 2016 - 10:50:30 EST


On 10/14/2016 6:30 PM, Mark Rutland wrote:
> On Fri, Oct 14, 2016 at 05:52:50PM +0100, Luis.Oliveira@xxxxxxxxxxxx wrote:
>> - is_slave = device_property_read_bool(&pdev->dev, "isslave");
> Which tree is this based on? I cant see the existing isslave property in
> mainline HEAD (commit 29fbff8698fc0ac1).

I'm replying for Luís, since at the moment he's not able to reply.

This is a custom property since this I2C core doesn't support both master and
slave at the same time.

>
>> +#ifndef CONFIG_ACPI
>> + is_slave = device_property_read_bool(&pdev->dev, "is-slave");
>> +#endif
> This ifdef is broken. At least for arm64, a single kernel image can be
> booted with either ACPI or DT. We need separate accessors for DT and
> ACPI to handle these differently, or you need to explicitly check
> whether or not you have ACPI or DT at runtime.

Thanks for the feedback, I'll take a look at this

>
> Thanks,
> Mark.

Thanks,
Ramiro